Biomathematics & Statistics Scotland
BioSS was involved in the analysis of the Human Proof of Principal Study (Human PPS), which was funded by the European Nutrigenomics Organisation (NuGO) and led by RINH scientists. This study produced the largest collection of high-dimensional omics data sets we have seen so far. The Human PPS tries to identify and characterise biological response in body fluids to a basic nutritional challenge. Ten subjects gave body fluid samples (blood, urine, saliva) at four time points after overnight fasting and an additional sample at a fifth time point after a 36 hour fasting challenge. Various metabolomic, proteomic and transcriptomic analyses were performed on the samples, creating a total of 19 high dimensional data sets.
BioSS developed a common analysis strategy for
these data sets, which included fitting linear mixed
models that allowed us to obtain an estimate of
the within-subject coefficient of variation (CV)
for each variable (metabolite, protein, gene) that
was measured. This analysis not only allows us to
compare the typical variability between technologies
and body fluids (see figure) but can also be used
in power/sample size calculations for future
experiments using similar technologies.
A boxplot of the coefficients of variation (CV) obtained from a linear mixed model reveals higher within-subject variability for saliva and platelet proteomics than for other data.
Further details from: Claus-Dieter Mayer
Article date 2009
